Simplifying ImageStats Code as a GSOC Project

[amohit94]

The existing ImagesStats application is basically implemented using open source libraries such as GWT, GXT, OpenCV, JavaCV, Vaadin. It has the following functionality:
 Authenticating NIH users using NIH iTrust.
 Fetching images from the PL Server which are stored in the PL Database.
 Search feature to search images on the Server using SOLR search syntax based on many criteria including
o Search by facet fields :Event Name, Ground truth status, Color Channels, Image dimensions
o Search other fields: Author of annotations, Image URL, Image id
 Annotate images
o Type of annotations: Face, Profile, and Skin.
o Draw annotations from FaceFinder.
o Manually draw rectangles (annotations) on the images.
o Save and restore annotations to DB.
o Two annotators (Initial/Final) can work to cross check
 Extract IplImage (defined in OpenCV) data structure:
 Export annotations and IplImage data structures to a local file.

I will be having the following tasks for the project:
Task 1: Remove the dependencies of the application on PL server on NLM
The first task is to remove all references of the PL server on the NLM and fetch images from the local file system and make the application functional. This is a major task and need to be finished before doing any other task.
Task 2: Implementing the alternative for PL SOLR
The PL SOLR is also a part of PL Server. In order to make the application more flexible and load images from different sources we need to create a system to search images from the different sources in order to replace the SOLR in the application. This system needs to be independent of the sources used to load the images.
Task 3: Replace MySQL by SQLite
SQLite is a light weight database system. The next task is to replace the MySQL used in the project by SQLite. MySQL in this application is used for storing the images annotations.
Task 4: Remove FaceFinder Technology
Since FaceFinder is not open source, it need to be removed from the project.
Task 5: Remove NIH iTrust
Implement authentication mechanism for the project on Apache Tomcat Server and replace the NIH iTrust for this mechanism with it. NIH iTrust is a component of the NLM site and needs to be removed to make the project a standalone application. In order to authenticate I’ll save user login data in the SQL Database and use Tomcat Framework to authenticate the users.
Task 6: Work on UI and further functionalities
Till this point the project is free of any dependencies and can be deployed on a Server by itself. The next part of the project is to work on the UI of the application in order to make it more interactive or any other additional functionality that needs to be implemented. The application which is presently there lacks any kind of UI. This task can focus on improving this aspect of the application.
One functionality that can be added in the application is to show annotations in thumbnails. This will include showing the annotations created on an image as thumbnails in a separate popup window.
Each subtask will be debugged and documented as soon as it gets completed. That means there will be a report submission roughly every week.
Timeline:
Before Coding Period Starts: Become more familiar with the libraries used in the application eg. GWT, GXT etc. Getting more familiar with the frame work of the application and all the prerequisites for the various tasks defined.
• Week 1: Start doing Task 1.
• Week 2: Complete Task 1
• Week 3: Start doing Task 2
• Week 4: Continuation of Task 2
• Week 5: Completion of Task 2.
• Week 6: Start doing Task 3.
• Week 7: Complete Task 3.
• Week 8: Complete Task 4
• Week 9: Start Task 5.
• Week 10: Complete Task 5.
• Week 11: Start Task 6.
• Week 12: Complete Task 6.
• Week 13: Finish up any small details of the project.
Please note any uncertainties or aspects which depend on further research or investigation.